Semantic Kernel

Cutting edge keyword extraction engine

Semantic Kernel is an entity linking engine with two crowns from leading international natural language competitions.
Studio Ousia has won two consecutive leading international natural language competitions.

The submission from the winning team (ousia) achieved surprisingly good performance on this difficult task, near the level of inter-rater agreement.

Noisy User-generated Text (W-NUT)

Team ousia outperformed all other participants, with a 69% performance increase with respect to the second ranked approach, the baseline system.

Named Entity rEcognition & Linking Challenge

Features

1. Keyword extraction with precision

Extracts entities from paragraphs with precision

2. Ambiguity resolution

Links the keywords individually with the entities from Wikipedia

3. Understanding the importance

Attaches the level of importance (helpfulness) to each keyword

4. Classification of keywords

Attaches DBpedia Ontology classes to each keyword

See technology page for technical details.

Price and Plan

Free Plan

  • Up to 200 requests per month
  • Maximum number of characters in one request is 10,000
  • Provided through Mashape (an API marketplace)

Standard Plan

  • Accepts 100,000 requests per month for 1,000USD
  • For additional requests, $0.01 per request
  • Maximum number of letters in one request is 10,000
  • Provided through Mashape (an API marketplace)
  • No automatic upgrade to premium plans. Contact us for larger volumes and customizations

Premium Plan

  • Customizable for larger volumes and additional functionalities
  • Contact us for details

Status for Customers

Follow us on Twitter.